Migraine, a prevalent neurological disorder, is characterized by recurrent episodes of severe headaches accompanied by debilitating symptoms such as nausea, photophobia and phonophobia. Its accurate diagnosis remains a challenge due to variability in clinical presentations and reliance on subjective assessments. This study proposes an advanced model based on supervised machine learning to classify different types of migraine. The methodology included five phases: data acquisition, preprocessing, evaluation of multiple classification algorithms (RF, DT, Gradient Boosting, Extreme Gradient Boosting, Adaptive Boosting, LR, Naive Bayes Gaussian, KNN and Support Vector Classifier), hyperparameter optimization and comprehensive evaluation. The RF algorithm obtained the best performance, reaching an accuracy of 92.9% and an average area under the curve (AUC) of 0.994, showing an outstanding ability to distinguish between different types of migraine. The analysis of the confusion matrix allowed the identification of classification patterns and areas for improvement, highlighting a high accuracy in majority classes and areas of confusion in minority classes. In addition, hyperparameter optimization improved the predictive capacity and robustness of the model. This work highlights the potential of machine learning to address the limitations of traditional clinical diagnosis, offering a scalable and affordable solution to improve migraine classification, especially in resource-limited settings. Future research will focus on the integration of heterogeneous data, temporal variables, and additional clinical features to increase the generalizability and applicability of the model in real clinical settings.